Abstract
[Objective] In order to promote the collaboration among researchers and the construction of academic groups, a hypergraph based recommendation algorithm SCRH is proposed for scientific collaboration in the field of stem cells.
[Methods] Firstly, the research collaboration hypernetwork based on Hypergraph structure is constructed, and then the structural similarity index and attribute similarity index are linearly fused. Finally, the research collaboration recommendation is realized through the similarity calculation between nodes.
[Results] On the collaboration recommendation task, the AUC and MR index values of SCRH are 0.88 and 2.35, which are increased by 14.3% and 25.2% respectively compared with the comparison algorithm.
[Limitations] SCRH only considers the author's text attribute in the node attribute similarity measurement, and does not make full use of the author's citation information, organization information, document level and other attribute information.
[Conclusions] SCRH also considers the structural and attribute characteristics of hypergraph, which can effectively complete the task of scientific research cooperation and recommendation in the field of stem cells.
|